Most enterprise leadership teams no longer need convincing that AI matters. The harder question is what happens between an ambitious AI strategy and something that works reliably inside the business.
Three numbers help explain the gap: 36, 85 and 14. They tell three parts of the story. 36 points to the foundation, clean and accessible data. 85 shows the operational improvement possible when that foundation is in place. And 14 highlights the readiness gap, how few organizations have built a unified customer view.
Start with 36 percent. In Salesforce research involving more than 2,000 agentic AI leaders, 36 percent of organizations seeing real returns identified clean, accessible data as the strongest predictor of success, tied with tightly scoped use cases.
That matches what we see in practice. Conversations often begin with the model or platform, but quickly return to data. If the organization cannot give the model reliable, governed and usable information, a more capable model does not remove the risk. It simply makes the problem move faster.
In one banking environment we worked with, customer information sat across five platforms, each using different identifiers. Reconstructing a customer’s full relationship could take days of manual effort. Before any agent could be trusted to act on that information, the underlying architecture had to be addressed, including the move toward a modern lakehouse and a semantic layer that allowed business concepts such as revenue or an at-risk account to mean the same thing across systems.
This is one of the less glamorous parts of AI transformation, but it is often the part that determines whether anything scales. The work starts with data definitions, ownership, access, quality and security before it reaches the model.
The second number is 85 percent. That is the reduction in underwriting turnaround time some AI-assisted credit platforms are reporting, taking processes that once required three to five business days down to a matter of hours.
What is interesting is not only the speed. It is how the work changes. One agent can retrieve bureau data, another can perform analysis, another can check compliance, and another can draft the outcome. The human underwriter is not removed. Instead, judgment is concentrated on the cases that genuinely need it.
For us, this is where Data & AI work becomes practical. The opportunity, the roadmap and the build cannot be treated as separate exercises. They have to connect, with the technology designed around the way decisions will actually be made. A technically successful pilot that cannot fit into the business process is still only a pilot.
The third number is 14. That is the share of organizations Gartner found had built a genuinely unified view of the customer. Customer 360 has been promised for decades, yet enterprise information still sits across structured systems and large volumes of unstructured content such as call transcripts, emails and service logs.
Agentic AI creates a new opportunity because it can read, synthesize and act across information that traditional systems struggled to use. But it only becomes valuable when the organization knows which data can be trusted, what an agent is allowed to do, and when a human needs to step in.
That is why governance cannot be an afterthought. Explainability, human override and drift monitoring are not simply compliance tasks. They are part of the operating infrastructure that allows AI to scale with confidence.
This is increasingly where our Data & AI teams spend their time with clients across APAC. Not choosing another model for the sake of it, but helping organizations get the foundations right: governed data, secure infrastructure, clear use cases and operating ownership.
At SoftwareOne, our Data & AI Services are built around that bridge. We start with data foundations, assessing maturity, modernizing platforms and putting governance and security in place, so analytics and AI have something reliable to stand on. From there, our analytics, AI and workplace AI teams help clients move from prioritized use cases into production, working across Microsoft, AWS and Google Cloud rather than locking them into a single stack. The goal is not a longer list of pilots, but AI that is owned, measured and embedded in how the business runs.
The sequence is straightforward, even if the work is not. Start with the right business problem. Build governance alongside the technology. Measure outcomes that matter to the business. And make sure someone is accountable for what happens once the system leaves the lab and becomes part of day-to-day operations.
AI ambition is already widespread. The differentiator now is execution. The organizations that turn that ambition into lasting value will be the ones that can point to a real change in how the business works, not simply another successful demonstration.
To learn more, visit us here.
Sponsored Links
About the Author
